[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Xen-devel] glitch in ethernet RX ring.

Ok, I ran into something interesting I think.

hardware setup: athlon xp 2000+ 1G ram 9 ide disks on 2 pci controllers
and onboard. Sis900 100Mbps FDX ethernet - onboard.

Domain-0: + debian patches + evms DM patches.
Domain-1: 2.4.27 + debian patches.

Last Changeset applied from xeno-unstable:
ChangeSet@xxxxxxxxxxx, 2004-09-02 18:33:39+01:00,

(yeah, it IS a bit behind by a few days. Took me a while to work out the
major bugs from the new debifying of the source. Looking promising now
though. More details later once I have the bugs out and my T1 line back
up. 8-P)

Issue: Upon starting Domain1, The xen scripts can't deal with the
existance of an eth0:0 alias interface. The xen-br0 will fail to be
completely configured, so I have to pmanually add IP route statements.

If I don't add the ip route for domain 1 after ~ 45 to 60 seconds I'll
get this: "bad buffer on RX ring!(-1)" And poof. Away goes xen. 8-P

If I do add the ip route via xen-br0, it never happens and is solid.

It sounds like Domain0 isn't droping unroutable packets fast enough and
is dying for any domain that has no route.

I'm updating the patchset today and will attempt to replicate this with
more detail and a serial console capture.

This SF.Net email is sponsored by BEA Weblogic Workshop
FREE Java Enterprise J2EE developer tools!
Get your free copy of BEA WebLogic Workshop 8.1 today.
Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.